“Gay son or thot daughter?” is usually treated as a vulgar preference-revelation exercise. This undersells it. The question is interesting because it forces the respondent to compare two very different kinds of variance in descendant outcomes while operating under severe informational compression.
\nThe first mistake is to treat the two options as symmetrical.
\n“Gay son” specifies a relatively narrow trait: a male child whose primary romantic and sexual attraction is toward men.
\n“Thot daughter” is not comparably precise. It ordinarily implies some bundle of high sociosexuality, conspicuous sexual self-presentation, multiple partners, weak conformity to conventional sexual norms, and possibly monetization of attention. Different speakers attach different weights to these components.
\nThis asymmetry matters.
\nOne option changes approximately one dimension of the child. The other potentially changes several.
\nU = W + R + G + S − C − V\n\nW = child welfare\nR = quality of the parent-child relationship\nG = expected grandchild value\nS = family social and reputational stability\nC = expected crisis-management cost\nV = variance penalty\nThe question is not whether homosexuality or promiscuity is “good” in the abstract. The question is which bundle produces the higher expected value conditional on everything else being held approximately constant.
\nThe strongest argument for Gay Son is variance reduction.
\nConditional on having a gay son, a large amount of his future life remains unspecified. He may be prudent or reckless, ambitious or lazy, monogamous or promiscuous, rich or poor, pleasant or irritating.
\nHis homosexuality alone does relatively little work in the model.
\nThis is important because people routinely smuggle unrelated traits into the gay-son branch. They imagine a specific personality, social milieu, or aesthetic presentation and then respond to that composite.
\nThat is analytically invalid.
\nIf the intervention is simply:
\nYour son is gay.\nthen the information content is modest.
\nThe largest conventional negative term is probably G, insofar as some respondents assign positive value to straightforward biological reproduction. A gay son has, on average, a less frictionless path to producing biological grandchildren with a partner than a heterosexual son.
\nBut this is a reduction in probability, not elimination.
\nGay men can and do become fathers. More importantly, the grandchild term should not be treated as infinite. A parent who places enormous weight on descendant count should state that preference explicitly rather than laundering it through generalized discomfort with homosexuality.
\nThere may also be social costs in environments where homosexuality is stigmatized, but these are environmental penalties imposed on the child rather than obvious intrinsic disutilities of the trait itself.
\nFrom a decision-theoretic standpoint, Gay Son is therefore unusually clean.
\nIt gives you one piece of information.
\nMost of the distribution remains intact.
\nThot Daughter is more complicated because the variable is poorly specified.
\nSuppose, for practical purposes, we define the condition as:
\nAn adult daughter with unusually high sociosexuality, conspicuous sexual self-presentation, and a strong revealed preference for short-term mating opportunities.\nThe first thing to notice is that people systematically overestimate the reproductive upside.
\nA daughter having more sexual partners does not imply proportionately more grandchildren.
\nModern contraception largely breaks this relationship.
\nIndeed, if high sociosexuality correlates with delayed family formation, unstable partnerships, or deliberate avoidance of pregnancy, the intuitive “at least she’ll give me grandkids” argument may be substantially weaker than advertised.
\nThe more important difference is variance.
\nA high-sociosexuality lifestyle potentially exposes the daughter to a broader distribution of outcomes involving partner quality, interpersonal conflict, unwanted pregnancy, sexual-health risk, reputational volatility, coercive relationships, and economically consequential decisions.
\nNone of these outcomes is guaranteed.
\nThe point is that the distribution is wider.
\nThis creates a parental risk-management problem.
\nThe relevant question is not:
\nDo I approve of my daughter having sex?\nThat is too coarse.
\nThe relevant question is:
\nDoes the specified behavioral profile increase the frequency of high-cost tail events enough to offset whatever benefits she derives from autonomy, pleasure, attention, relationships, and optionality?\nI think the answer is probably yes under the ordinary meaning of “thot.”
\nThis is not because female sexuality is intrinsically costly.
\nIt is because “thot” is doing far more semantic work than “gay.”
\nPeople are often reluctant to include S, reputational effects, because this sounds morally unserious.
\nThat is a mistake.
\nSocial reputation has material consequences whether or not those consequences are fair.
\nIf a behavior predictably alters employment opportunities, relationship markets, family conflict, social trust, or exposure to harassment, those consequences belong in the model.
\nSaying “people should not judge her” does not make judgment disappear.
\nSimilarly, saying “people should not discriminate against my gay son” does not make homophobia disappear.
\nBoth branches therefore contain externally imposed social costs.
\nThe relevant comparison is empirical.
\nMy expectation is that in most contemporary Western professional environments, the expected reputational penalty attached to being gay is now considerably narrower than the penalty associated with highly conspicuous sexual behavior.
\nThis varies enormously by subculture.
\nA gay son in rural Mississippi and a thot daughter in Brooklyn do not inhabit the same social environment as a gay son in Brooklyn and a thot daughter in rural Mississippi.
\nThe question is underspecified unless location and class are fixed.
\nI think a large fraction of male respondents are covertly solving:
\nmax P(grandchildren)\nrather than:
\nmax U(child’s life)\nThis explains several otherwise strange answers.
\nIf the daughter’s heterosexual activity is treated as direct evidence of reproductive potential, some respondents will tolerate enormous variance elsewhere in the model.
\nThis is probably a mistake.
\nA grandchild has positive utility to many parents, but it is not obvious that one additional expected grandchild dominates large changes in the adult child’s welfare, stability, or relationship quality.
\nThere is also a hidden asymmetry.
\nPeople often imagine the gay son remaining permanently childless while imagining the thot daughter eventually settling down and producing a conventional family.
\nThis inserts mean reversion into one branch but not the other.
\nIf you allow the thot daughter to become a monogamous 34-year-old mother of two, you should also allow the gay son to become a married 34-year-old father of two.
\nOtherwise the comparison is rigged.
\nThere is one term that respondents tend to underweight because it is difficult to quantify:
\nR\nThe child eventually learns what kind of parent you are.
\nIf your preference between these two outcomes is strong enough to affect how you treat the child, that response itself creates utility consequences.
\nA father who selects Gay Son but spends twenty years punishing his daughter for sexual behavior may have chosen the wrong branch in a way the original hypothetical does not capture.
\nLikewise, a father who selects Thot Daughter but would make a gay son feel fundamentally defective has created a large negative term unrelated to the underlying trait.
\nThis means the chooser’s own psychology should be considered endogenous.
\nYou are not choosing between two children in a vacuum.
\nYou are choosing between two children who will be raised by the person making the choice.
\nUnder ordinary interpretations, I choose Gay Son.
\nThe reason is not that homosexuality is intrinsically preferable to female promiscuity.
\nThe reason is that the Gay Son intervention is narrower, lower-variance, and less behaviorally predictive.
\n“Gay” tells me who my son is attracted to.
\n“Thot” tells me something about what my daughter repeatedly does.
\nBehavioral variables generally carry more downstream information than orientation variables.
\nIf the hypothetical were instead:
\nGay son or heterosexual daughter who has many consensual sexual partners, uses contraception reliably, avoids coercive relationships, is psychologically healthy, and experiences no meaningful negative consequences from her sex life?\nthen much of the difference disappears.
\nBut that is not the conventional question.
\nThe conventional question contains its answer partly in the amount of information packed into each label.
\nGay Son is one variable.
\nThot Daughter is a latent-factor model disguised as slang.
\n","excerpt":"“Gay son or thot daughter?” is usually treated as a vulgar preference-revelation exercise. This undersells it. The question is interesting because it forces the respondent to compare two very different kinds of variance in descendant outcomes while operating under severe informational compression.","tags":["Decision Theory","Family","Applied Ethics","Risk Analysis"],"score":561,"displayOrder":30,"thumbnail":"/assets/articles/gay-son-or-thot-daughter-thumb.webp","image":"/assets/articles/gay-son-or-thot-daughter.webp","commentCount":18,"date":"2026-07-23T00:00:00.000Z","updatedAt":"2026-09-22T18:57:01.044Z","revision":3,"status":"published"},"comments":[{"id":"seed-gay-son-or-thot-daughter-0","author":"BayesBender","text":"Strong upvote. Most of the disagreement I've seen about this question comes from people silently defining \"thot\" as either \"woman who has sex\" or \"an unbroken sequence of maximally adverse tail events,\" depending on which answer they already wanted to give.","parentId":null,"score":184,"date":"2026-07-23T00:50:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-1","author":"latent_variable","text":"The two labels differ in type. Orientation is a trait. \"Thot\" is already a hostile behavioral summary written by somebody else. You can get nearly any result you like by putting an information-rich insult in one branch and a neutral demographic category in the other.","parentId":null,"score":153,"date":"2026-07-23T01:30:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-3","author":"sociosexual_inventory","text":"If you're going to operationalize \"thot\" as high sociosexuality you should know the SOI-R has three subscales (behavior, attitude, desire) and none of them measures self-presentation. Your definition mixes behavior with presentation, and those come apart a lot in practice. Plenty of people score high on behavior and post nothing, and plenty post constantly and score low on all three. You'd want self-presentation as its own construct, at which point \"thot daughter\" is at least a two-factor model and probably three.","parentId":null,"score":97,"date":"2026-07-23T02:15:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-2","author":"variance_minimizer","text":"Agreed that the asymmetry is embedded in the conventional wording. That asymmetry is my central claim. I can't remove it without changing the hypothetical into a different and less popular one.","parentId":"seed-gay-son-or-thot-daughter-1","score":68,"date":"2026-07-23T02:40:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-5","author":"mean_reversion_mike","text":"The rigged-comparison section is the best part of the post but you don't follow it all the way through. You allow mean reversion in the grandchildren term and then evaluate the variance term as if she stays 22 forever. The 34-year-old mother of two you mention is a low-variance person. Integrate V over her whole life and the gap shrinks a lot, possibly enough to flip it for parents with long horizons.","parentId":null,"score":112,"date":"2026-07-23T03:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-4","author":"variance_minimizer","text":"That supports the post's conclusion. The more factors the label loads on, the more information it carries relative to \"gay.\"","parentId":"seed-gay-son-or-thot-daughter-3","score":41,"date":"2026-07-23T03:30:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-6","author":"variance_minimizer","text":"Fair. I evaluated at the age the question implicitly refers to. I've added a note on time integration. The ranking holds for any discount rate I'd consider reasonable, though the margin narrows for parents who expect to live past 85.","parentId":"seed-gay-son-or-thot-daughter-5","score":52,"date":"2026-07-23T04:45:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-7","author":"lauren_t_s","text":"As a former thot by the post's definition (roughly 2014 to 2019, I'm happy to be a data point), I'd ask you to separate variance experienced by the child from variance experienced by the parent. Most of the variance in my case showed up in my parents. My outcomes were fine. Better than fine in a few respects, I met most of my current friends then and I learned a lot about reading people quickly, which turned out to be useful at work.\n\nWhat my parents experienced was five years of not knowing where I was on Saturday nights, and that was costly for them. But it's a cost to the chooser, and the post presents V as if it's about the child's welfare.","parentId":null,"score":134,"date":"2026-07-23T05:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-8","author":"variance_minimizer","text":"That's a useful distinction. In my model V is the parent's variance penalty, since the parent is the one making the choice. I'll make that explicit.","parentId":"seed-gay-son-or-thot-daughter-7","score":47,"date":"2026-07-23T06:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-9","author":"lauren_t_s","text":"Then W and V are measuring different people's welfare and you should probably not add them.","parentId":"seed-gay-son-or-thot-daughter-8","score":88,"date":"2026-07-23T06:50:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-10","author":"eldercare_economist","text":"Nobody's mentioned the term that settles it for most of the parents I know, which is who looks after you at 80. In my extended family it has been the gay son every single time, three for three. Small sample, I know. But if that pattern is anywhere close to general it should dominate everything else in the model.","parentId":null,"score":66,"date":"2026-07-23T07:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-11","author":"variance_minimizer","text":"This would enter through R. I'm reluctant to make the choice conditional on which child is more likely to become an unpaid caregiver, but I agree it moves the margin toward Gay Son.","parentId":"seed-gay-son-or-thot-daughter-10","score":23,"date":"2026-07-23T08:20:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-12","author":"dlc_poll_results","text":"I ran this as a poll at our meetup in June. 31 respondents, 24 chose gay son. Of the 7 who chose thot daughter, 6 had no children, which I take as evidence the question measures something different for people who are answering it hypothetically.","parentId":null,"score":58,"date":"2026-07-23T10:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-13","author":"tupelo_rationalist","text":"For what it's worth I live in rural Mississippi. Your 2x2 is roughly right but the gay son penalty here is smaller than people in Brooklyn assume. Everyone has a cousin. The thot daughter penalty is about what you'd expect, mainly because everyone also knows everyone's mother.","parentId":null,"score":39,"date":"2026-07-23T12:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-14","author":"fecundity_max","text":"I'll defend max P(grandchildren) since nobody else is going to. I have four kids and I'd answer this purely on expected grandchildren. Your contraception point is right as far as it goes but you're ignoring the upside tail. Some fraction of those partners turn into marriages, and you don't need many.","parentId":null,"score":31,"date":"2026-07-24T01:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-15","author":"mean_reversion_mike","text":"That's mean reversion again, just with a different name.","parentId":"seed-gay-son-or-thot-daughter-14","score":19,"date":"2026-07-24T03:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-16","author":"wagmi_utilitarian","text":"gay son is cope. thot daughter has a higher expected income and you all know it","parentId":null,"score":-21,"date":"2026-07-24T08:00:00.000Z","own":false,"moderatable":false,"seed":true},{"id":"seed-gay-son-or-thot-daughter-17","author":"lurker_since_2019","text":"Coming back to this because my daughter turned 13 last week. I would like to withdraw my upvote and replace it with a question about whether anyone has a version of this model for the next five years specifically.","parentId":null,"score":27,"date":"2026-09-25T00:00:00.000Z","own":false,"moderatable":false,"seed":true}]}